2026-04-21 14:59:41.193512: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776783581.203642 1486732 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776783581.206710 1486732 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776783581.216680 1486732 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776783581.216692 1486732 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776783581.216694 1486732 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776783581.216696 1486732 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 14:59:44.601991: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0421 14:59:44.719185 137667750730880 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0421 15:00:11.714420 137667750730880 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 15:00:11.714699 137667750730880 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 15:00:11.714745 137667750730880 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0421 15:00:11.714776 137667750730880 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0421 15:00:11.714956 137667750730880 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 15:00:12.072243 137667750730880 maxtext_utils.py:1551] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 15:00:12.072617 137667750730880 checkpointing.py:677] Setting up checkpoint logger... I0421 15:00:12.072664 137667750730880 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 15:00:12.072753 137667750730880 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 15:00:12.073601 137667750730880 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7d34fbbaef90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 15:00:14.485435 137667750730880 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 15:00:14.485706 137667750730880 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d33ab9ac440>}, handler_registry=None I0421 15:00:14.486118 137667750730880 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d33ab9ac440>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0421 15:00:14.486163 137667750730880 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3353d72360>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0421 15:00:14.486195 137667750730880 composite_checkpoint_handler.py:505] Initialized registry DefaultCheckpointHandlerRegistry({('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d33ab9ac440>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d33ab9ac440>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3353d72360>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3353d72360>}). I0421 15:00:14.486485 137667750730880 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 15:00:14.486548 137667750730880 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7d3353d94ea0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 15:00:14.624446 137667750730880 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints I0421 15:00:14.624732 137667750730880 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d335412bb60> I0421 15:00:14.624822 137667750730880 checkpointing.py:301] Checkpoint manager created! I0421 15:00:14.979768 137667750730880 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0421 15:00:14.982756 137667750730880 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0421 15:00:15.035708 137667750730880 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0421 15:00:15.058263 137667750730880 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0421 15:00:15.058324 137667750730880 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0421 15:00:15.834470 137667750730880 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 15:00:15.834578 137667750730880 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:00:16.136610 137667750730880 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:00:16.136700 137667750730880 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:00:16.149196 137667750730880 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:00:16.149246 137667750730880 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:00:16.198225 137667750730880 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:00:16.198303 137667750730880 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:00:16.210861 137667750730880 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:00:16.210912 137667750730880 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:00:16.223306 137667750730880 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:00:16.223356 137667750730880 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:00:16.243221 137667750730880 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 15:00:16.243282 137667750730880 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:00:16.292016 137667750730880 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 15:00:16.292101 137667750730880 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 15:00:16.671359 137667750730880 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 15:00:16.671462 137667750730880 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0421 15:00:19.157315 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157424 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157463 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157494 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157521 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157546 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157568 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157605 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157628 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157650 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157673 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157694 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157716 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157736 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157759 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157780 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157801 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157821 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157842 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157862 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157882 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157903 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157923 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157944 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.157964 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.157988 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158009 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158029 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158066 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158089 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158111 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158132 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158152 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158172 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158192 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158213 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158234 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158253 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158272 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158292 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158310 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0421 15:00:19.158330 137667750730880 maxtext_utils.py:1654] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0421 15:00:19.158369 137667750730880 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 15:00:19.158411 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 15:00:19.158438 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 15:00:19.158474 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0421 15:00:19.158509 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 15:00:19.158533 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 15:00:19.158569 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0421 15:00:19.158605 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0421 15:00:19.158631 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0421 15:00:19.158656 137667750730880 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0421 15:00:19.158689 137667750730880 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 15:00:19.158721 137667750730880 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 15:00:20.612164 137667750730880 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:00:20.612254 137667750730880 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:00:20.624160 137667750730880 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:00:20.624209 137667750730880 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:00:31.496861 137667750730880 max_utils.py:791] Total memory size: 3.5 GB, Output size: 1.5 GB, Temp size: 1.9 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0421 15:00:31.497781 137667750730880 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 15:00:31.498142 137667750730880 metric_logger.py:301] number parameters: 1.105 billion I0421 15:00:46.858436 137667750730880 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0421 15:00:46.994486 137667750730880 checkpointing.py:776] Waited 0.13604164123535156 seconds for step 0 to finish before starting checkpointing. I0421 15:00:46.995197 137667750730880 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:00:46.995382 137667750730880 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0421 15:00:46.995875 137667750730880 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0. I0421 15:00:47.078405 137667750730880 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0421 15:00:47.170129 137667750730880 jax_array_handlers.py:347] Scheduling D2H of 81 prioritized jax.Array. I0421 15:00:47.170228 137667750730880 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0421 15:00:47.191158 137592634869312 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0 I0421 15:00:47.883015 137576147060288 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0/items I0421 15:00:48.293237 137576111408704 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776783647793743925, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776783647793743925, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0421 15:00:48.742649 137667750730880 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.573031s I0421 15:00:48.743345 137667750730880 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 7.421 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0421 15:00:48.743412 137667750730880 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.663123s (batch_requests_ready=0.082669s, total_serialization_initiated=1.579875s, others=0.000580s) I0421 15:00:48.743489 137667750730880 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.664517s (all_items=0.000029s, per_item={'items': '0.00002885'}, temp_paths=1.664489) I0421 15:00:48.747600 1490092 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 15:00:48.747661 1490092 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 15:00:48.751716 137592645355072 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:00:48.751871 137667750730880 async_checkpointer.py:561] Finished blocking save. Time taken: 1.756444s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0. I0421 15:00:48.752272 137667750730880 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:00:48.752525 137576006551104 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:00:48.752611 137667750730880 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776783646.995181, 'wait_for_prev_duration_secs': 4.553794860839844e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776783646.995403, 'checkpointer_blocking_duration_secs': 1.7567837238311768, 'get_old_steps_start_time': 1776783648.7522037, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776783646.9950716, 'checkpoint_manager_blocking_duration_secs': 1.7575099468231201} I0421 15:00:48.752825 137667750730880 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0421 15:00:48.752876 137667750730880 max_utils.py:750] Memstats: After params initialized: I0421 15:00:48.752926 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_0(process=0,(0,0,0,0)) I0421 15:00:48.752952 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_1(process=0,(1,0,0,0)) I0421 15:00:48.752973 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_2(process=0,(0,1,0,0)) I0421 15:00:48.752993 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_3(process=0,(1,1,0,0)) I0421 15:00:48.753011 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_4(process=0,(0,2,0,0)) I0421 15:00:48.753030 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_5(process=0,(1,2,0,0)) I0421 15:00:48.753064 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_6(process=0,(0,3,0,0)) I0421 15:00:48.753083 137667750730880 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_7(process=0,(1,3,0,0)) I0421 15:00:49.201029 137667750730880 metric_logger.py:196] completed step: 0, seconds: 15.359, TFLOP/s/device: 0.885, Tokens/s/device: 133.341, total_weights: 13328, loss: 10.844, lm_loss: 10.844, perplexity: 51212.484 I0421 15:00:49.202012 137667750730880 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/tensorboard/' I0421 15:00:49.309316 137667750730880 metric_logger.py:196] completed step: 1, seconds: 2.339, TFLOP/s/device: 5.809, Tokens/s/device: 875.607, total_weights: 12332, loss: 10.871, lm_loss: 10.871, perplexity: 52634.344 I0421 15:00:49.421015 137667750730880 metric_logger.py:196] completed step: 2, seconds: 0.013, TFLOP/s/device: 1071.540, Tokens/s/device: 161514.196, total_weights: 15161, loss: 10.871, lm_loss: 10.871, perplexity: 52617.816 I0421 15:00:49.520977 137576058979904 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0/items/array_metadatas/process_0 I0421 15:00:49.533076 137667750730880 metric_logger.py:196] completed step: 3, seconds: 0.109, TFLOP/s/device: 124.218, Tokens/s/device: 18723.544, total_weights: 13327, loss: 10.857, lm_loss: 10.857, perplexity: 51888.457 I0421 15:01:08.046643 137667750730880 metric_logger.py:196] completed step: 4, seconds: 0.112, TFLOP/s/device: 121.743, Tokens/s/device: 18350.432, total_weights: 11939, loss: 10.861, lm_loss: 10.861, perplexity: 52094.562 I0421 15:01:08.105448 137667750730880 metric_logger.py:196] completed step: 5, seconds: 0.113, TFLOP/s/device: 119.915, Tokens/s/device: 18074.947, total_weights: 15502, loss: 10.882, lm_loss: 10.882, perplexity: 53230.324 I0421 15:01:08.214578 137667750730880 metric_logger.py:196] completed step: 6, seconds: 18.558, TFLOP/s/device: 0.732, Tokens/s/device: 110.355, total_weights: 13864, loss: 10.886, lm_loss: 10.886, perplexity: 53423.605 I0421 15:01:08.326361 137667750730880 metric_logger.py:196] completed step: 7, seconds: 0.011, TFLOP/s/device: 1287.269, Tokens/s/device: 194031.265, total_weights: 12988, loss: 10.829, lm_loss: 10.829, perplexity: 50459.859 I0421 15:01:08.438308 137667750730880 metric_logger.py:196] completed step: 8, seconds: 0.109, TFLOP/s/device: 124.250, Tokens/s/device: 18728.339, total_weights: 13820, loss: 10.848, lm_loss: 10.848, perplexity: 51456.266 I0421 15:01:08.445412 137667750730880 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0421 15:01:08.662227 137667750730880 checkpointing.py:776] Waited 0.21676993370056152 seconds for step 10 to finish before starting checkpointing. I0421 15:01:08.662838 137667750730880 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:01:37.508379 137576048494144 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 48.764421s (commit=48.204008s, array_metadata_write=0.560413s) I0421 15:01:37.509821 137592645355072 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 250.609 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 50 seconds) (per-host) I0421 15:01:37.509944 137592645355072 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 48.757873s. I0421 15:01:37.720478 137592645355072 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776783647793743925, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0421 15:01:37.908188 137592645355072 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:01:38.091599 137576111408704 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776783647793743925, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0421 15:01:38.312862 137592645355072 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:01:38.313916 137592645355072 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.553438s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0/items I0421 15:01:38.314568 137592645355072 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0/items I0421 15:01:38.553628 137592645355072 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0 I0421 15:01:39.234396 137592645355072 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0`. I0421 15:01:39.235181 137592645355072 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 52.239762s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/0 I0421 15:01:39.235260 137592645355072 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 50.483191s. I0421 15:01:39.235401 137576006551104 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:01:39.235453 137576006551104 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:01:39.235510 137576006551104 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:01:39.235551 137576006551104 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0421 15:01:39.235715 137667750730880 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0421 15:01:39.235873 137667750730880 checkpoint_manager.py:1441] Waiting for previous save to complete took 30.573076 seconds. If this number is high, consider checkpointing less frequently. I0421 15:01:39.236894 137667750730880 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0421 15:01:39.237236 137667750730880 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10. I0421 15:01:39.422878 137667750730880 jax_array_handlers.py:347] Scheduling D2H of 81 prioritized jax.Array. I0421 15:01:39.422982 137667750730880 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0421 15:01:39.426280 137576006551104 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10 I0421 15:01:40.157326 137576027522624 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10/items I0421 15:01:49.154186 137667750730880 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.731769s I0421 15:01:49.161110 137667750730880 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.256 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0421 15:01:49.161375 137667750730880 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.827983s (batch_requests_ready=0.083119s, total_serialization_initiated=9.737860s, others=0.007004s) I0421 15:01:49.161494 137667750730880 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.828572s (all_items=0.000021s, per_item={'items': '0.00002074'}, temp_paths=9.828551) I0421 15:01:49.162966 137576058979904 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:01:49.163090 137667750730880 async_checkpointer.py:561] Finished blocking save. Time taken: 9.926148s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10. I0421 15:01:49.163288 137667750730880 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:01:49.163439 137592645355072 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:01:49.163522 137667750730880 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776783668.6627612, 'wait_for_prev_duration_secs': 30.573076486587524, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776783699.236918, 'checkpointer_blocking_duration_secs': 9.926286458969116, 'get_old_steps_start_time': 1776783709.1632185, 'get_old_steps_duration_secs': 3.170967102050781e-05, 'checkpoint_manager_blocking_start_time': 1776783668.662682, 'checkpoint_manager_blocking_duration_secs': 40.50081491470337} I0421 15:01:49.163620 137667750730880 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0421 15:01:49.164353 137667750730880 metric_logger.py:196] completed step: 9, seconds: 0.111, TFLOP/s/device: 121.975, Tokens/s/device: 18385.356, total_weights: 12300, loss: 10.857, lm_loss: 10.857, perplexity: 51900.086 I0421 15:01:49.174750 137667750730880 metric_logger.py:196] completed step: 10, seconds: 0.111, TFLOP/s/device: 122.336, Tokens/s/device: 18439.818, total_weights: 13436, loss: 10.876, lm_loss: 10.876, perplexity: 52905.383 I0421 15:01:49.283769 137667750730880 metric_logger.py:196] completed step: 11, seconds: 40.727, TFLOP/s/device: 0.334, Tokens/s/device: 50.287, total_weights: 13817, loss: 10.853, lm_loss: 10.853, perplexity: 51714.605 I0421 15:01:49.395645 137667750730880 metric_logger.py:196] completed step: 12, seconds: 0.009, TFLOP/s/device: 1575.319, Tokens/s/device: 237449.275, total_weights: 12076, loss: 10.852, lm_loss: 10.852, perplexity: 51622.902 I0421 15:01:49.507590 137667750730880 metric_logger.py:196] completed step: 13, seconds: 0.111, TFLOP/s/device: 122.869, Tokens/s/device: 18520.193, total_weights: 11048, loss: 10.864, lm_loss: 10.864, perplexity: 52261.953 I0421 15:01:49.619496 137667750730880 metric_logger.py:196] completed step: 14, seconds: 0.112, TFLOP/s/device: 121.380, Tokens/s/device: 18295.679, total_weights: 14267, loss: 10.834, lm_loss: 10.834, perplexity: 50738.680 I0421 15:01:49.731234 137667750730880 metric_logger.py:196] completed step: 15, seconds: 0.112, TFLOP/s/device: 121.496, Tokens/s/device: 18313.184, total_weights: 13004, loss: 10.832, lm_loss: 10.832, perplexity: 50634.074 I0421 15:01:49.843220 137667750730880 metric_logger.py:196] completed step: 16, seconds: 0.112, TFLOP/s/device: 121.540, Tokens/s/device: 18319.901, total_weights: 13811, loss: 10.867, lm_loss: 10.867, perplexity: 52422.980 I0421 15:01:49.955184 137667750730880 metric_logger.py:196] completed step: 17, seconds: 0.112, TFLOP/s/device: 121.599, Tokens/s/device: 18328.754, total_weights: 13709, loss: 10.850, lm_loss: 10.850, perplexity: 51525.906 I0421 15:01:49.967865 137576090437184 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10/items/array_metadatas/process_0 I0421 15:01:50.070076 137667750730880 metric_logger.py:196] completed step: 18, seconds: 0.112, TFLOP/s/device: 121.385, Tokens/s/device: 18296.496, total_weights: 12831, loss: 10.839, lm_loss: 10.839, perplexity: 50989.945 I0421 15:01:50.178673 137667750730880 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0421 15:01:50.180301 137667750730880 checkpointing.py:776] Waited 0.0016384124755859375 seconds for step 19 to finish before starting checkpointing. I0421 15:01:50.180774 137667750730880 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:02:13.368979 137576079951424 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 24.207358s (commit=23.772748s, array_metadata_write=0.434610s) I0421 15:02:13.370284 137576058979904 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 371.307 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host) I0421 15:02:13.370405 137576058979904 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 24.207265s. I0421 15:02:13.799348 137576058979904 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:02:14.196809 137576058979904 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:02:14.197627 137576058979904 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.545917s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10/items I0421 15:02:14.198414 137576058979904 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10/items I0421 15:02:14.444505 137576058979904 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10 I0421 15:02:15.126626 137576058979904 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10`. I0421 15:02:15.127367 137576058979904 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.890434s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/10 I0421 15:02:15.127449 137576058979904 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 25.964313s. I0421 15:02:15.127570 137592645355072 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:02:15.127620 137592645355072 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:02:15.127665 137592645355072 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:02:15.127702 137592645355072 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 15:02:15.127863 137667750730880 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0421 15:02:15.127999 137667750730880 checkpoint_manager.py:1441] Waiting for previous save to complete took 24.947238 seconds. If this number is high, consider checkpointing less frequently. I0421 15:02:15.128236 137667750730880 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0421 15:02:15.128553 137667750730880 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19. I0421 15:02:15.311680 137667750730880 jax_array_handlers.py:347] Scheduling D2H of 81 prioritized jax.Array. I0421 15:02:15.311781 137667750730880 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0421 15:02:15.314436 137576058979904 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19 I0421 15:02:16.054673 137576048494144 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19/items I0421 15:02:27.349489 137667750730880 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 12.038201s I0421 15:02:27.356651 137667750730880 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.017 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host) I0421 15:02:27.356723 137667750730880 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 12.140081s (batch_requests_ready=0.088659s, total_serialization_initiated=12.044393s, others=0.007030s) I0421 15:02:27.356808 137667750730880 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 12.140683s (all_items=0.000023s, per_item={'items': '0.00002337'}, temp_paths=12.140660) I0421 15:02:27.358445 137576079951424 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:02:27.358535 137667750730880 async_checkpointer.py:561] Finished blocking save. Time taken: 12.230253s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19. I0421 15:02:27.358703 137667750730880 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:02:27.358930 137592645355072 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:02:27.359071 137667750730880 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776783710.1807451, 'wait_for_prev_duration_secs': 24.947238445281982, 'time_between_consecutive_saves_sec': 10.945178985595703, 'checkpointer_blocking_start_time': 1776783735.1282606, 'checkpointer_blocking_duration_secs': 12.230378866195679, 'get_old_steps_start_time': 1776783747.358655, 'get_old_steps_duration_secs': 1.7404556274414062e-05, 'checkpoint_manager_blocking_start_time': 1776783710.1807046, 'checkpoint_manager_blocking_duration_secs': 37.17832899093628} I0421 15:02:27.359188 137667750730880 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0421 15:02:27.359220 137667750730880 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:02:28.089152 137576069465664 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19/items/array_metadatas/process_0 I0421 15:02:52.641520 137576048494144 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 25.284020s (commit=24.867537s, array_metadata_write=0.416483s) I0421 15:02:52.642772 137576079951424 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 337.681 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0421 15:02:52.642827 137576079951424 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 25.284250s. I0421 15:02:53.067460 137576079951424 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:02:53.517567 137576079951424 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:02:53.518411 137576079951424 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.585097s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19/items I0421 15:02:53.519169 137576079951424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19/items I0421 15:02:53.759097 137576079951424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19 I0421 15:02:54.405671 137576079951424 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19`. I0421 15:02:54.406421 137576079951424 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 39.278147s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_05_fp8/checkpoints/19 I0421 15:02:54.406489 137576079951424 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 27.047912s. I0421 15:02:54.406706 137592645355072 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:02:54.406831 137592645355072 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:02:54.406886 137592645355072 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:02:54.406930 137592645355072 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 15:02:54.407129 137667750730880 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0421 15:02:54.407292 137667750730880 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:02:54.409276 137667750730880 metric_logger.py:196] completed step: 19, seconds: 0.116, TFLOP/s/device: 117.552, Tokens/s/device: 17718.715, total_weights: 13717, loss: 10.851, lm_loss: 10.851, perplexity: 51591.992 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops